Regression-based risk estimation
Injects domain knowledge via empirical rules — active operation at lunch/dinner, consistent ignition cycles, gaps in peak-time operation, etc. as key variables.
AI Risk Scoring
We score a rider's future with multimodal data.
BCM's core technology asset. Combining operating data with an automotive-bred team, we predict each rider's credit (non-payment) and accident (insurance) risk in advance.
AI RISK SCORE
Bond risk, accident risk, multimodal data
Designed so external evaluators grasp that a quantitative model exists from the first screen.
Four multimodal data types
Beyond automotive FMS's single driving stream, we integrate a motorcycle rider's daily life into four data types.
| Data type | Source | Key items |
|---|---|---|
| Driving · vehicle | FMS device (embedded) | Distance, ignition cycles, time-of-day, idle ratio, hard acceleration/braking, model/year, IMU acceleration, impact, GPS |
| Activity | Smartphone (activity) | Travel distance, step count, delivery patterns, peak-time operation |
| Contract · operations | Internal ERP | Maintenance adherence, payment method (auto-debit), deposit level, delinquency history, rental history |
| Biometric · behavioral | Smartphone (bio/behavioral) | Heart-rate variability, fatigue, stress level, rest patterns |
Rule + Learning hybrid model
Multidimensional anomaly detection
Autoencoder · LSTM Encoder · temporal CNN → Isolation Forest · LOF · K-means/DBSCAN → Logistic Regression · XGBoost to quantify accident/non-payment probability.
Validation data (per the TIPS project)
Credit risk — 85-rider analysis
| Group | Count | Delinquency |
|---|---|---|
| Low (low risk) | 13 | 0.0% |
| Mid | 41 | 4.9% |
| High (high risk) | 31 | 16.1% |
In the High group, 4 of 5 showed concentrated, repeated delinquency compared with the Low group's cases (38.5%).
Target performance (TIPS evaluation items)
- Credit-risk prediction AUC ≥ 0.65
- Accident-risk prediction AUC ≥ 0.65
- With long-term data accumulation (goal) 0.70+
- e-Call false-positive rate < 15%